@devo-bmad-custom/agent-orchestration 1.0.0

package/src/.agents/skills/audit-website/references/OUTPUT-FORMAT.md

@@ -0,0 +1,250 @@
+# LLM Format Output Reference
+
+## Overview
+
+The `--format llm` output is a compact, token-optimized hybrid XML/text format designed specifically for AI agent consumption. It provides structured audit data in a format that balances machine readability with token efficiency.
+
+## Key Characteristics
+
+- **40-70% smaller** than verbose XML format
+- **1-space indentation** for minimal token usage
+- **Hybrid structure**: XML tags + text prefixes for metadata
+- **Inline attributes**: Metadata stored as XML attributes, not nested elements
+- **Comma-separated lists**: Pages and arrays formatted inline
+- **Flattened hierarchy**: Reduced nesting depth compared to verbose XML
+
+## Format Structure
+
+### 1. Document Header
+
+```xml
+<?xml version="1.0" encoding="UTF-8"?>
+<audit version="0.0.24">
+```
+
+### 2. Site Information
+
+```xml
+<site url="https://example.com" crawled="51" date="2025-01-18T10:30:00Z"/>
+```
+
+Attributes:
+- `url` - Base URL audited
+- `crawled` - Number of pages crawled
+- `date` - ISO 8601 timestamp
+
+### 3. Health Score
+
+```xml
+<score overall="85" grade="B">
+ <cat name="Core SEO" score="92"/>
+ <cat name="Technical SEO" score="88"/>
+ <cat name="Content Quality" score="76"/>
+</score>
+```
+
+Attributes:
+- `overall` - 0-100 health score
+- `grade` - Letter grade (A-F)
+- Categories with individual scores
+
+### 4. Summary
+
+```xml
+<summary passed="120" warnings="15" failed="8"/>
+```
+
+Attributes:
+- `passed` - Number of passed checks
+- `warnings` - Number of warnings
+- `failed` - Number of failed checks
+
+### 5. Issues Section
+
+Issues are grouped by category with compact inline metadata:
+
+```xml
+<issues>
+ <category name="Core SEO" errors="2" warnings="3">
+  <rule id="core/meta-title" severity="error" status="fail">
+   Missing or empty meta title tags
+   Desc: Every page should have a unique meta title
+   Fix: Add descriptive <title> tags to each page
+   Pages (2): https://example.com/about, https://example.com/contact
+   Items (2):
+    - https://example.com/about
+    - https://example.com/contact
+  </rule>
+  <rule id="core/meta-description" severity="warning" status="warn">
+   Desc: Pages should have meta descriptions
+   Fix: Add <meta name="description"> tags
+   Pages (5): https://example.com/page1, https://example.com/page2, ...
+  </rule>
+ </category>
+ <category name="Performance" errors="0" warnings="1">
+  ...
+ </category>
+</issues>
+```
+
+#### Rule Structure
+
+Each `<rule>` element contains:
+
+**Attributes:**
+- `id` - Rule identifier (e.g., `core/meta-title`)
+- `severity` - `error`, `warning`, or `info`
+- `status` - `pass`, `warn`, or `fail`
+
+**Text Content (in order):**
+1. **Message** (optional) - Human-readable issue summary
+2. **Desc:** - Rule description (what's being checked)
+3. **Fix:** - Recommended solution (how to fix)
+4. **Pages (n):** - Comma-separated list of affected URLs
+5. **Items (n):** - Dash-prefixed list of specific items with metadata
+
+### 6. Items Format
+
+Items provide detailed context about affected elements:
+
+```xml
+Items (3):
+ - https://example.com/missing-title (title: "")
+ - https://example.com/duplicate-title (title: "Home Page") (from: https://example.com/other)
+ - /broken-link [status: 404, type: internal] (from: https://example.com/contact)
+```
+
+Item format:
+- `- <id>` - Primary identifier (URL, selector, etc.)
+- `(<label>)` - Optional human-readable label if different from id
+- `[key: value, ...]` - Metadata in square brackets
+- `(from: <sources>)` - Source pages where item appears
+
+## Diff Output (LLM Format)
+
+When using `squirrel report --diff` or `--regression-since` with `--format llm`,
+the output is a compact XML diff format:
+
+```xml
+<?xml version="1.0" encoding="UTF-8"?>
+<diff version="0.0.24">
+ <baseline id="a7b3c2d1" url="https://example.com" date="2026-01-17T10:30:00Z" pages="42" score="87" grade="B"/>
+ <current id="b9c4e1f2" url="https://example.com" date="2026-01-18T10:30:00Z" pages="44" score="84" grade="B"/>
+ <summary added="3" removed="1" changed="2" regressions="1" improvements="1"/>
+ <added>
+  <issue fp="abc123" rule="core/meta-title" severity="error" status="fail" check="meta-title" category="core" weight="8">
+   Missing page title
+   Target: page /about
+  </issue>
+ </added>
+ <removed>
+  ...
+ </removed>
+ <changed>
+  <change type="regression" fp="def456" rule="links/broken-links" severity="warning" status="fail" check="broken-links">
+   warn→fail: Broken link count increased
+   Before:
+    <issue ...> ... </issue>
+   After:
+    <issue ...> ... </issue>
+  </change>
+ </changed>
+</diff>
+```
+
+Key fields:
+- `fp`: deterministic fingerprint for the issue instance
+- `rule`, `check`, `severity`, `status`: rule and check metadata
+- `Target:`: item/page/check target
+- `change type`: `regression`, `improvement`, or `change`
+
+## Example Output
+
+```xml
+<?xml version="1.0" encoding="UTF-8"?>
+<audit version="0.0.24">
+<site url="https://example.com" crawled="51" date="2025-01-18T10:30:00Z"/>
+<score overall="78" grade="C">
+ <cat name="Core SEO" score="85"/>
+ <cat name="Technical SEO" score="92"/>
+ <cat name="Performance" score="65"/>
+</score>
+<summary passed="98" warnings="12" failed="5"/>
+<issues>
+ <category name="Core SEO" errors="2" warnings="1">
+  <rule id="core/meta-title" severity="error" status="fail">
+   Missing meta title on 2 pages
+   Desc: Every page should have a unique meta title
+   Fix: Add descriptive <title> tags to each page
+   Pages (2): https://example.com/about, https://example.com/contact
+   Items (2):
+    - https://example.com/about
+    - https://example.com/contact
+  </rule>
+ </category>
+</issues>
+</audit>
+```
+
+## Usage
+
+The LLM format is available via both `audit` and `report` commands:
+
+```bash
+# Direct LLM output (single step)
+squirrel audit https://example.com --format llm
+
+# Or two-step workflow
+squirrel audit https://example.com
+squirrel report <audit-id> --format llm
+
+# Pipe directly to AI agent
+squirrel audit https://example.com --format llm | claude
+```
+
+The `audit` command supports `--format llm` directly for convenience. Use the two-step workflow when you need to generate reports in multiple formats from a single audit.
+
+## Comparison with Other Formats
+
+| Format | Size | Structure | Best For |
+|--------|------|-----------|----------|
+| `xml` | 209KB | Verbose, 2-space indent, fully nested | Enterprise integration, archival |
+| `llm` | 125KB | Compact, 1-space indent, hybrid | AI agents, token-limited contexts |
+| `json` | 180KB | Structured data | Programmatic processing |
+| `text` | 45KB | Plain text, no structure | Simple piping, grep |
+
+## Token Efficiency
+
+The LLM format achieves 40-70% size reduction compared to verbose XML through:
+
+1. **1-space indentation** instead of 2-4 spaces
+2. **Inline attributes** instead of nested elements
+3. **Text prefixes** (Desc:, Fix:) instead of XML tags
+4. **Comma-separated lists** instead of multiple elements
+5. **Flattened hierarchy** - fewer nesting levels
+
+## XML Character Escaping
+
+Special characters are properly escaped:
+- `&` → `&amp;`
+- `<` → `&lt;`
+- `>` → `&gt;`
+- `"` → `&quot;`
+- `'` → `&apos;`
+
+## Design Philosophy
+
+The LLM format is optimized for:
+1. **Token efficiency** - Critical for API cost and context limits
+2. **Easy parsing** - XML structure for reliable extraction
+3. **Human readability** - AI agents can explain issues naturally
+4. **Progressive disclosure** - Summary → Categories → Rules → Items
+5. **Actionable insights** - Fix recommendations included inline
+
+## Implementation Notes
+
+- Generated by `generateLlmReport()` in `app/src/reports/output/llm.ts`
+- Empty issues section renders as self-closing: `<issues/>`
+- All text content is XML-escaped for safety
+- Indentation uses spaces only (no tabs)
+- Line endings are Unix-style (`\n`)

package/src/.agents/skills/clean-code-standards/SKILL.md

@@ -0,0 +1,105 @@
+---
+name: clean-code-standards
+description: DRY/SOLID code quality principles for architectural review — 23 rules across 7 categories. Use when performing a DRY/SOLID review (DR/DRY trigger) or when architect-agent is running a combined AR+PMR+DRY review gate.
+version: 1.0.0
+author: bmad
+tags: [code-quality, solid, dry, architecture, review]
+---
+
+# Clean Code Standards
+
+## How to Use This Skill
+
+Load this skill when performing a DRY/SOLID architectural review. For each file in the target area, evaluate against all 23 rules below. Produce findings with ID format `DRY-001`, `DRY-002`, etc.
+
+When running as the combined **AR+PMR+DRY gate sub-agent** (architect-agent), run this skill as the third lens after AR and PMR. Produce a separate findings block appended to the gate output.
+
+## Categories & Rules
+
+### SOLID Principles
+
+- **solid-srp** — Single Responsibility Principle: each class/module has one reason to change. Flag classes that mix data access, business logic, and presentation.
+- **solid-ocp** — Open/Closed Principle: open for extension, closed for modification. Flag code that requires internal changes to add new behaviour (missing abstractions, hardcoded switch/if-else on type).
+- **solid-lsp** — Liskov Substitution Principle: subtypes must be substitutable for base types without altering correctness. Flag overrides that change base-class contracts or raise new exceptions.
+- **solid-isp** — Interface Segregation Principle: no client should depend on interfaces it doesn't use. Flag large interfaces/protocols forcing classes to implement irrelevant methods.
+- **solid-dip** — Dependency Inversion Principle: depend on abstractions, not concretions. Flag direct instantiation of concrete dependencies inside high-level modules (should be injected).
+
+### Core Principles
+
+- **core-dry** — Don't Repeat Yourself: every piece of knowledge has a single, authoritative representation. Flag 3+ near-identical code blocks; flag duplicated constants, query patterns, or validation logic.
+- **core-kiss** — Keep It Simple, Stupid: prefer the simplest solution that works. Flag over-engineered abstractions, unnecessary indirection, or patterns applied where they add no value.
+- **core-yagni** — You Aren't Gonna Need It: don't add functionality until it's needed. Flag speculative generality (hooks/plugin systems for hypothetical future needs, unused configuration options).
+- **core-soc** — Separation of Concerns: divide the program into distinct sections, each addressing a separate concern. Flag components that mix routing + business logic + data formatting in one function.
+- **core-lod** — Law of Demeter: a unit should have only limited knowledge of other units (`a.b.c.d()` is a violation). Flag deep chained property access and tight coupling across module boundaries.
+- **core-fail-fast** — Fail Fast: detect and report errors at the earliest possible point, not silently. Flag swallowed exceptions, empty catch blocks, and silent fallbacks that hide bugs.
+- **core-encapsulation** — Encapsulation: hide implementation details, expose only what's necessary. Flag public fields that should be properties, direct model field access across module boundaries.
+
+### Design Patterns
+
+- **pattern-repository** — Repository Pattern: abstract data access behind a consistent interface. Flag direct ORM calls scattered across business logic or routes instead of going through a repository.
+- **pattern-factory** — Factory Pattern: use factories for complex object construction. Flag constructor overload or multi-step object building repeated in multiple places.
+- **pattern-observer** — Observer Pattern: use events/callbacks over tight coupling. Flag modules that directly call many other modules on state change (should emit events/signals instead).
+
+### Code Organization
+
+- **org-module-size** — Modules should be small and focused. Flag files over 300 lines — strong signal that SRP is violated and the module should be split.
+- **org-function-size** — Functions should do one thing. Flag functions over 30 lines — likely mixing multiple concerns. Flag functions that return different types depending on control flow.
+- **org-nesting** — Avoid deep nesting (3+ levels). Flag deeply nested if/for/try blocks — prefer early returns and extracted helpers.
+
+### Naming
+
+- **naming-intent** — Names should reveal intent. Flag abbreviations (`usr`, `fn`, `tmp`), single-letter variables outside loops (`x`, `d`), and boolean names that don't read as questions (`active` instead of `is_active`).
+- **naming-consistency** — Consistent naming conventions across the codebase. Flag mixed conventions in the same module (e.g., camelCase and snake_case Python, or inconsistent REST verb usage in route names).
+
+### Functions
+
+- **func-args** — Functions should have ≤3 parameters. Flag functions with 4+ parameters — use a config object/dataclass. Exception: well-known constructors with established signatures.
+- **func-side-effects** — Functions should not have hidden side effects. Flag functions whose names imply read-only behaviour but also write to DB, cache, or emit events without declaring it.
+
+### Documentation
+
+- **doc-why-not-what** — Comments should explain WHY, not WHAT the code does. Flag comments that merely restate the code (`# increment counter` above `counter += 1`). Flag missing WHY comments on non-obvious business rule implementations.
+
+---
+
+## Severity Guide
+
+| Severity | Condition | Examples |
+|---|---|---|
+| 🔴 Critical | SOLID violations causing tight coupling or untestable code; DRY violations with 3+ duplicated blocks of ≥10 lines | God class violating SRP, direct DB calls scattered across routes, copy-pasted 50-line validation blocks |
+| 🟡 Major | Core principle violations; module/function size violations; naming issues that reduce readability | 60-line function, 400-line module, YAGNI speculation, Law of Demeter chain, missing fail-fast |
+| 🟢 Minor | Documentation, minor style, isolated single-instance patterns | Missing WHY comment, one slightly long variable name, single unused config option |
+
+---
+
+## Finding Format
+
+```markdown
+### [DRY-001] {title}
+File: {path}:{line}
+Rule: {rule-id} — {rule name}
+Issue: {description of the violation}
+Fix: {specific fix — include corrected code snippet where helpful}
+```
+
+---
+
+## Status Rules
+
+- `passed` — zero critical findings and ≤3 major findings
+- `needs_fixes` — any critical finding OR 4+ major findings
+
+---
+
+## YAML Verdict (return to review-orchestrator or Krakken)
+
+```yaml
+dry_review_verdict:
+  target: {path or feature name}
+  actionable_count: {N}
+  critical: {count of 🔴}
+  major: {count of 🟡}
+  minor: {count of 🟢}
+  findings_path: _bmad-output/features/{feature-slug}/planning/dry-review-findings-{artifact_id}.md
+  status: passed | needs_fixes
+```

package/src/.agents/skills/excalidraw-dark-standard/SKILL.md

@@ -0,0 +1,282 @@
+---
+name: excalidraw-dark-standard
+description: Squid-Master dark mode Excalidraw design standard — colors, shapes, typography, spacing, and naming convention for all system diagrams
+version: 1.0.0
+author: bmad
+tags: [diagrams, excalidraw, design-system, dark-mode, visualization]
+---
+
+# Excalidraw Dark Mode Design Standard
+
+## Overview
+
+All Squid-Master Excalidraw diagrams follow this standard. Load this skill before creating or editing any `.excalidraw` file. It defines the full visual language: canvas setup, semantic colors, shape vocabulary, typography, spacing, and file naming.
+
+Use the `excalidraw-diagram-generator` skill alongside this one — it provides the JSON generation engine. This skill provides the Squid-Master color and structure rules that override its defaults.
+
+---
+
+## Canvas Setup
+
+Every diagram MUST start with a massive background rectangle as the **first element** in the `elements` array:
+
+```json
+{
+  "type": "rectangle",
+  "x": -5000,
+  "y": -5000,
+  "width": 10000,
+  "height": 10000,
+  "backgroundColor": "#1e1e2e",
+  "strokeColor": "#1e1e2e",
+  "fillStyle": "solid",
+  "strokeWidth": 1,
+  "roughness": 0,
+  "opacity": 100
+}
+```
+
+Excalidraw `appState`:
+```json
+{
+  "theme": "dark",
+  "viewBackgroundColor": "#1e1e2e",
+  "gridSize": null,
+  "exportWithDarkMode": true
+}
+```
+
+---
+
+## Color Palette
+
+### Background Colors
+
+| Token | Hex | Use |
+|---|---|---|
+| `canvas-bg` | `#1e1e2e` | Canvas background fill rect + appState |
+| `deep-bg` | `#0f0f1a` | Section separators, nested container interiors |
+| `subtle-border` | `#2a2a3e` | Card outlines, frame borders, dividers |
+
+### Element Fill Colors (semantic)
+
+| Token | Hex | Use |
+|---|---|---|
+| `fill-agent` | `#1e3a5f` | Agent nodes — primary actors in a workflow |
+| `fill-skill` | `#1a4d2e` | Skill nodes — tools loaded by agents |
+| `fill-orchestrator` | `#2d1b69` | Orchestrator / gate / coordinator nodes |
+| `fill-error` | `#4a2020` | Halt, error, escalation states |
+| `fill-loop` | `#3d2a00` | Loop counters, retry logic, autonomous pass steps |
+| `fill-user` | `#1a3040` | User approval gates, human-in-the-loop steps |
+| `fill-annotation` | `#252525` | Notes, metadata, file path labels |
+| `fill-track` | transparent | Track / group container — use stroke only |
+
+### Stroke / Accent Colors
+
+| Token | Hex | Use |
+|---|---|---|
+| `stroke-agent` | `#4a9eed` | Agent borders, primary data-flow arrows |
+| `stroke-skill` | `#22c55e` | Skill borders, pass badges |
+| `stroke-orchestrator` | `#a78bfa` | Gate borders, concurrent group frames |
+| `stroke-error` | `#f87171` | Halt borders, error states |
+| `stroke-loop` | `#fbbf24` | Loop borders, retry/warning states |
+| `stroke-user` | `#38bdf8` | User action borders |
+| `stroke-track` | `#7dd3fc` | Track/section frame borders (dashed) |
+| `text-primary` | `#e5e5e5` | All node labels |
+| `text-muted` | `#888888` | Annotations, metadata, sublabels |
+
+---
+
+## Shape Vocabulary
+
+### Rectangle — Agent
+```
+fill: #1e3a5f  ·  stroke: #4a9eed  ·  strokeWidth: 2  ·  borderRadius: 6
+size: 180×60px (standard)  ·  240×60px (wide, for orchestrators)
+```
+Use for: every named agent (`architect-agent`, `security-agent`, `dev-agent`, etc.)
+
+### Rectangle — Skill
+```
+fill: #1a4d2e  ·  stroke: #22c55e  ·  strokeWidth: 2  ·  borderRadius: 6
+size: 180×50px
+```
+Use for: skills loaded by an agent (`clean-code-standards`, `security-review`, `ui-ux-pro-custom`)
+
+### Rectangle — Orchestrator / Gate
+```
+fill: #2d1b69  ·  stroke: #a78bfa  ·  strokeWidth: 2  ·  borderRadius: 6
+size: 240×60px
+```
+Use for: `review-orchestrator`, gate labels, coordination hubs
+
+### Rectangle — Halt / Error
+```
+fill: #4a2020  ·  stroke: #f87171  ·  strokeWidth: 2  ·  borderRadius: 6
+```
+Use for: `🛑 HALT`, escalation exits, max-loop failure states
+
+### Rectangle — Loop / Step
+```
+fill: #3d2a00  ·  stroke: #fbbf24  ·  strokeWidth: 2  ·  borderRadius: 6
+```
+Use for: loop pass counters (`Pass N of max`), retry steps, autonomous repeat nodes
+
+### Ellipse — User Action
+```
+fill: #1a3040  ·  stroke: #38bdf8  ·  strokeWidth: 2
+size: 160×60px
+```
+Use for: `👤 USER APPROVAL`, human confirmation gates
+
+### Diamond — Decision
+```
+fill: #2d1b69  ·  stroke: #a78bfa  ·  strokeWidth: 2
+size: 120×80px
+```
+Use for: `passed / needs_fixes`, `loop count?`, `halt or continue?`, branch points
+
+### Frame / Group (dashed container)
+```
+fill: transparent  ·  stroke: #a78bfa  ·  strokeStyle: dashed  ·  strokeWidth: 1
+```
+Use for: concurrent sub-agent groups, track boundaries, review gate containers. Label in top-left corner at `text-muted`.
+
+### Note / Annotation
+```
+fill: #252525  ·  stroke: #555555  ·  strokeStyle: dashed  ·  strokeWidth: 1  ·  borderRadius: 4
+```
+Use for: file path labels (`dry-review-findings-{artifact_id}.md`), session state fields, context metadata
+
+### Arrow — Data / Control Flow
+```
+stroke: #4a9eed  ·  strokeWidth: 2  ·  endArrowhead: arrow
+```
+- **Solid** — primary control or data flow
+- **Dashed** (`strokeStyle: dashed`) — optional, conditional, or fallback paths
+- Label arrows with `text-muted` (12px) when the relationship needs clarification
+
+---
+
+## Typography
+
+**Rule:** All text uses `fontFamily: 5` (Nunito). No exceptions.
+
+| Role | Size | Weight | Color |
+|---|---|---|---|
+| Diagram title | 24px | 700 | `#a78bfa` |
+| Section / track label | 18px | 600 | `#7dd3fc` |
+| Node label (primary) | 14px | 600 | `#e5e5e5` |
+| Node sublabel | 12px | 400 | `#aaaaaa` |
+| Annotation / metadata | 11px | 400 | `#666666` |
+| Arrow label | 12px | 400 | `#888888` |
+
+In Excalidraw JSON:
+```json
+{
+  "type": "text",
+  "text": "architect-agent",
+  "fontSize": 14,
+  "fontFamily": 5,
+  "textAlign": "center",
+  "verticalAlign": "middle",
+  "strokeColor": "#e5e5e5"
+}
+```
+
+---
+
+## Spacing Rules
+
+| Property | Value | Notes |
+|---|---|---|
+| Horizontal gap between siblings | 200–300px | Node edge to node edge |
+| Vertical gap between rows | 100–150px | Node edge to node edge |
+| Standard node size | 180×60px | Agents, skills, steps |
+| Wide node size | 240×60px | Orchestrators, gates |
+| Frame padding | 40px | Inset from outermost child node |
+| Group label offset | 8px from top-left corner | Inside frame, above first child |
+| Stroke width (nodes) | 2px | 1px for annotations |
+| Border radius | 6px | All rectangles |
+| Arrow strokeWidth | 2px | |
+
+---
+
+## Status Badge Labels (inline in diagram)
+
+Add small rounded-rectangle badges next to verdict outputs:
+
+| Badge | Fill | Stroke | Text |
+|---|---|---|---|
+| `✅ passed` | `#1a4d2e` | `#22c55e` | `#22c55e` |
+| `❌ needs_fixes` | `#4a2020` | `#f87171` | `#f87171` |
+| `🔁 LOOP` | `#3d2a00` | `#fbbf24` | `#fbbf24` |
+| `🛑 HALT` | `#4a2020` | `#f87171` | `#f87171` |
+| `🟣 CONCURRENT` | `#2d1b69` | `#a78bfa` | `#a78bfa` |
+| `👤 USER APPROVAL` | `#1a3040` | `#38bdf8` | `#38bdf8` |
+
+Badge size: auto-width × 24px, font 11px, border-radius 12px.
+
+---
+
+## File Naming Convention
+
+```
+{scope}-{subject}[-{variant}].excalidraw
+```
+
+### Scopes
+
+| Prefix | Use |
+|---|---|
+| `sm-` | Squid-Master system diagrams |
+| `sq-` | Squidhub platform diagrams |
+| `bd-` | BMAD agent framework diagrams |
+| `feat-` | Feature-specific diagrams |
+
+### Standard Diagram Names
+
+| Diagram | Filename |
+|---|---|
+| All tracks overview | `sm-tracks-overview.excalidraw` |
+| Agent + skills flow | `sm-agents-skills.excalidraw` |
+| All workflows overview | `sm-workflows-overview.excalidraw` |
+| instructions.md visual map | `sm-instructions-map.excalidraw` |
+| Review track detail | `sm-review-track.excalidraw` |
+
+### Storage
+
+- **New diagrams:** `_bmad-output/diagrams/`
+- **Legacy / archived:** `_bmad-output/diagrams/artifacts/`
+
+---
+
+## Diagram Structure Checklist
+
+Before writing any `.excalidraw` file:
+
+1. [ ] Background rect is first element — `#1e1e2e`, 10000×10000, x/y −5000
+2. [ ] `appState.theme = "dark"`, `viewBackgroundColor = "#1e1e2e"`
+3. [ ] All text uses `fontFamily: 5`
+4. [ ] Every node fill and stroke matches semantic color table
+5. [ ] Spacing: 200–300px horizontal, 100–150px vertical
+6. [ ] Concurrent groups wrapped in dashed `#a78bfa` frame
+7. [ ] Diagram title present (24px, `#a78bfa`, top-left)
+8. [ ] Section labels present for multi-section diagrams
+9. [ ] Filename follows `{scope}-{subject}.excalidraw` convention
+10. [ ] Saved to `_bmad-output/diagrams/` (not `artifacts/`)
+
+---
+
+## How to Use This Skill with excalidraw-diagram-generator
+
+1. Load this skill (`excalidraw-dark-standard`) — establishes color/shape rules
+2. Load `excalidraw-diagram-generator` — provides JSON structure and generation patterns
+3. Describe the diagram content in terms of the shape vocabulary above
+4. Generate elements array using generator's patterns, applying THIS skill's colors
+5. Always prepend the background rect before all other elements
+6. Write output to `_bmad-output/diagrams/{filename}.excalidraw`
+
+---
+
+_v1.0 — 2026-03-14_